Mont Terri Project

Transcription

Mont Terri Project

ANDRA BGR CRIEPI ENRESA GRS HSK IRSN
JAEA NAGRA OBAYASHI SCK•CEN SWISSTOPO
Mont Terri Project
TECHNICAL REPORT 2007-05
(TN 2007-45)
November 2007
LT Experiment:
Strength and Deformation of Opalinus Clay
W. Gräsle and I. Plischke
Federal Institute for Geosciences and Natural
Resources (BGR), Germany
Mont Terri Project, TR 2007-05
Distribution:
Standard distribution:
ANDRA (J. Delay)
BGR (H. J. Alheid)
CRIEPI (K. Kiho)
ENRESA (J. Astudillo)
GRS (T. Rothfuchs)
HSK (E. Frank)
IRSN (J.-M. Matray)
JAEA (N. Shigeta)
Nagra (M. Hugi)
Obayashi (H. Kawamura, T. Tanaka)
SCKyCEN (G. Volckaert)
SWISSTOPO (P. Bossart and P. Hayoz)
GI AG (Ch. Nussbaum)
Additional distribution:
Every organisation & contractor takes care of their own distribution.
LT-Exp. Phase 11+12
TABLE OF CONTENTS
ABSTRACT ........................................................................................................................
1
1
CONTEXT AND AIM OF THE STUDY ..............................................................................
1
2
SAMPLING AND PREPARATION ....................................................................................
2
3
GEOLOGICAL AND MINERALOGICAL DESCRIPTION .................................................
4
4
DRAINED CREEP TEST ....................................................................................................
5
4.1
CONCEPT AND OBJECTIVE ................................................................................................
5
4.2
RESULTS .........................................................................................................................
9
COMPLEX STRENGTH TEST ........................................................................................... 13
5
5.1
5.1.1
5.1.2
5.1.3
CONCEPT AND OBJECTIVE ................................................................................................
Test section A – limit of linear elasticity ...................................................................
Test section B – shear strength ...............................................................................
Test section C – residual strength ...........................................................................
13
13
19
19
5.2
5.2.1
5.2.2
5.2.3
RESULTS .........................................................................................................................
Test section A – limit of linear elasticity ...................................................................
Test section B – shear strength ...............................................................................
Test section C – residual strength ...........................................................................
19
22
24
25
6
SUMMARY AND PERSPECTIVE ...................................................................................... 26
7
REFERENCES ................................................................................................................... 26
8
LIST OF TABLES .............................................................................................................. 27
9
LIST OF FIGURES ............................................................................................................. 27
LT-Exp. Phase 11+12
Page 1
ABSTRACT
Long-term multistep creep tests on samples from Mont Terri BLT-14 and BLT-15 boreholes were carried out. The tests were performed at 30°C under drained conditions. Sample orientations parallel
(BLT-14) as well as perpendicular (BLT-15) to the bedding plane were investigated to account for the
anisotropy of the material.
A new type of triaxial test using a complex process path was developed to yield as much information
on the mechanical behavior as possible from a single sample. Relatively large data sets not affected
by the variations between different samples are considered to facilitate the derivation of a constitutive
equation for Opalinus clay. A first test of the new experimental layout was carried out on a BHE-B1
sample, investigating the limit of linear elastic behavior, the shear failure limit, and the residual
strength as functions of minimum principal stress. The test was performed under undrained conditions
to allow for the investigation of possible pore pressure effects.
1 CONTEXT AND AIM OF THE STUDY
The long-term safety of mines and repositories for radioactive or toxic wastes can be predicted if the
mechanical behavior and hydraulic properties of the host rock can be described reliably. For this purpose laboratory tests were performed to investigate the deformation of the host rock at all relevant
stress and temperature conditions.
At the Mont Terri test site in Switzerland the Opalinus clay stone is investigated for its suitability for the
disposal of radioactive waste. The LT-experiment (Laboratory Testing) is focused on the investigation
of mechanic and hydraulic material properties and behavior (in particular creep, dilatation, healing,
strength, permeability and anisotropy) in a relevant range of stress states and temperatures. The tests
are performed in the laboratories of BGR.
Like all bedded rocks, Opalinus clay has different mechanical properties in and perpendicular to its
foliation plane i.e. orthorhombic symmetry with three mutually perpendicular planes of symmetry.
BGR’s laboratory program will describe and define all the rock mechanic properties of the Opalinus
clay parallel as well as orthogonal to the bedding plane (two-dimensional orthotropic solid). We intend
especially to investigate the deformation processes, the distortion-dilatation relation, and to obtain a
constitutive equation for the Opalinus clay from the Mont Terri site.
Page 2
LT-Exp. Phase 11+12
2 SAMPLING AND PREPARATION
Several drilling campaigns were accomplished by BGR during phase 11+12 to obtain rock samples for
the ongoing and planned tests. The boreholes BLT-14 to BLT-18 are located close to the Northern end
of Gallery 98 in the shaly facies of the Opalinus clay at the Mont Terri site (Fig. 1). The starting points
of the boreholes are in a distance of 8 to 10 m from the boundary between Opalinus clay and Jurensis
marls.
BHE-B1
BLT-15+16
BLT-17+18
BLT-14
Fig. 1 Location of the BLT-14 to BLT-18 boreholes drilled during phase 11+12, and of the BHE-B1
borehole used in the complex strength test.
In January 2007 borehole BLT-14 was drilled parallel to the bedding within strike. End of March 2007
BLT-15 was drilled perpendicular to the bedding, followed by BLT-16 parallel to the bedding within fall.
BLT-15 went into Jurensis marls at a depth of approximately 8.2 m.
Finally, BLT-17 and BLT-18 were drilled in July 2007. BLT-17 which has an angle of 60° towards the
bedding normal to strike and is almost parallel to the security gallery was abandoned because all
cores broke apart. The vertical, axially aligned interfaces were obviously freshly formed fractures
(Fig. 2). Considering that BLT-17 is located in the zone of increased stresses slightly outside the EDZ
of the security gallery, these fractures may most likely result from the quick unloading of the material
during the drilling process. Thus, the attempt to gain cores in this alignment was not repeated at this
location. Instead, BLT-18 was drilled 30° towards the bedding normal to strike.
Opalinus clay material is very susceptible to damage by desiccation (SCHNIER, 2004). Furthermore,
the impact of oxygen significantly affects the material properties. This is mainly attributed to the oxidation of pyrite and the subsequent formation of gypsum.
To minimize these perturbations, the samples were sealed in air-tight aluminum coated foils as quick
as possible. Transport and storage of the sealed samples is done in special liners (Fig. 3), filled with
nitrogen as a protective gas at a pressure of 3 bar. This will keep the protected conditions even in
case of a perforation of the sealing foil.
LT-Exp. Phase 11+12
Page 3
Fig. 2 Core from BLT-17 showing freshly formed axial fractures. There is neither a slickenside (which
would indicate a tectonic fault) nor any trace of gypsum crystals (which generate within a few days in
any crack opened in the EDZ due to the impact of atmospheric oxygen) on the interface.
Page 4
LT-Exp. Phase 11+12
Fig. 3 The liners used for transport and storage of core samples.
Sample preparation was carried out immediately before the tests. They were cut to length with a band
saw, and then they were trimmed to smooth cylindrical shape on a lathe. To avoid undesirable friction
effects during the tests the samples were covered with a Teflon foil before they were coated in rubber
tubes which prevent any contact between the samples and the oil pressurizing the Karman cell.
3 GEOLOGICAL AND MINERALOGICAL DESCRIPTION
The general mineralogical properties of the host rocks at the Mont Terri laboratory have been summarized by THURY & BOSSART (1999). Shaly, calcareous-sandy and sandy facies are distinguished. Their
main difference in terms of mineralogical composition is the content of clay minerals and quartz. The
content of feldspars is similar (see also PEARSON ET AL. 2003). All samples used for the rock mechanical experiments covered by this report belong to the shaly facies.
Detailed investigations of the mineralogical composition of the rocks from BLT-14 and BLT-15 are in
work. First results show that the carbonate contents of the sample from BLT-14/05 is 19.2 wt-% and of
BLT-14/11 15.5 wt.-%. At BLT-14/05 the carbonate is located as calcite crystals in zones, whereas at
BLT-14/11 carbonate exists as detritus in small layers. More microstructural investigations of Opalinus
clay from Mont Terri especially of the distribution of carbonate are shown at KLINKENBERG ET AL.
(2007). Pictures of all tested samples are in Fig. 5 and Fig. 6.
LT-Exp. Phase 11+12
Page 5
4 DRAINED CREEP TEST
4.1 CONCEPT AND OBJECTIVE
The long term behavior of rock material under the
impact of anisotropic stresses is a key parameter for
the evolution of any underground excavation (convergence, reduction of deviatoric stress). Therefore,
creep tests where carried out on Mont Terri Opalinus
clay. The anisotropy of the clay has to be considered
in these tests. As the pore pressure is expected to
influence the creep behavior of Opalinus clay a well
defined hydraulic regime has to be kept during the
tests.
The creep tests were carried out in conventional Karman cells. Four samples measuring approx. 100 mm
in diameter and 220 mm in length were investigated.
Two samples from BLT-14 were drilled parallel to the
bedding plane (p-samples); the others from BLT-15
were drilled normal to bedding (s-samples, Fig. 4).
Fig. 5 and Fig. 6 display the prepared samples before
the tests. A summary of the tested samples and the
test procedure is given in Tab. 1.
Fig. 4 The alignment of cylindrical
samples with respect to the bedding
for p- and s-geometry.
Tab. 1 Sample characterization and test layout for the drained creep tests.
sample
BLT-14/05
BLT-14/11
BLT-15/4
BLT-15/8
07001
07002
07004
07005
diameter [mm]
99.9
99.9
99.8
99.9
length [mm]
220.0
220.0
218.2
220.7
bulk density [g/cm3]
2.448
2.439
2.444
2.447
mass [kg]
4.222
4.207
4.171
4.232
velocity p-wave [m/s]
3320
3325
N.T.2)
2698
velocity s-wave [m/s]
N.T.
2)
2)
2)
orientation to bedding
file-number
1)
depth [m]
duration of phases [d]
cell pressure σx [MPa]
1)
2)
σz – σx = 0.5 MPa
N.T.
N.T.
N.T.2)
p
p
s
s
3.00 – 3.22
9.74 – 9.96
4.53 – 4.75
7.37 – 7.60
10
10
10
10
2
2
1
1
σz – σx = 1
MPa
17
17
17
17
σz – σx = 5
MPa
36
36
38
38
σz – σx = 8
MPa
50
50
–
–
σz – σx = 10
MPa
79
79
52
52
σz – σx = 13
MPa
69
69
69
69
The file-number identifies a sample in the BGR laboratory documentation system LIMS.
N.T. = no transmission of signal.
Page 6
LT-Exp. Phase 11+12
LT-Exp. Phase 11+12
Fig. 5 The p-geometry samples BLT-14/05 and BLT-14/11 before the creep test.
Page 7
Page 8
LT-Exp. Phase 11+12
LT-Exp. Phase 11+12
Page 9
Fig. 6 The s-geometry samples BLT-15/04 and BLT-15/08 before the creep test.
At both ends a disk of sintered metal was placed between the sample and the pressure plate. This
provides an additional pore volume to absorb water that might leak from the sample, thus avoiding a
persistent build up of pore pressure. Due to the pronounced anisotropy of hydraulic conductivity in a
laminated material like Opalinus clay, this should work well for p-samples, whereas in case of
s-samples, a pore pressure induced by loading the sample might persist quite a while until it dies
down as a consequence of hydraulic flow.
All tests were performed under a constant cell pressure σx = 1 MPa. After 17 days of compaction at an
axial load of σz = 2 MPa, creep was investigated in a multi-step loading scheme applying axial stresses
of 6, 9, 11 and 14 MPa. (Note the sign convention used throughout this report: Compressive stresses
and strains are defined as positive, extensive stresses and strains as negative.) Each level lasted at
least 30 days. According to sample availability the tests on BLT-15 samples were started three month
later than those from BLT-14.
4.2 RESULTS
The observed deformations of the four samples are shown in Fig. 7. Overall, the deformations are very
small and none of the samples exhibits a significantly increasing deformation for deviatoric stresses
σdev = σ1 – σ3 up to 10 MPa (Fehler! Verweisquelle konnte nicht gefunden werden.). Thus, no creep
Page 10
LT-Exp. Phase 11+12
could be found in these tests for σdev ≤ 10 MPa. Surprisingly, indications of an increase in sample
length are found in several cases (yellow shading in Fehler! Verweisquelle konnte nicht gefunden
werden.), two of them even showing correlations with R2 > 0.6 (BLT-15/04 at σdev = 5 MPa, BLT-14/05
at σdev = 8 MPa). There is no temperature drift that might explain these paradoxial observation.
Therefore, it is most likely attributable to some drift of the displacement measurement system.
There are indications of creep in three of the four samples for a deviatoric stress of σdev = 13 MPa
(Fehler! Verweisquelle konnte nicht gefunden werden., Fig. 8). But these deformations still do not
exceed the preceeding fluctuations significantly. In particular, the observed rates of deformation are of
the same order as the largest "paradoxical extension rates" mentioned above. Thus, the calculated
creep rates dε/dt include a high level of uncertainty. Nevertheless, the dε/dt-values are clearly below
the values reported from various studies (compiled in CZAIKOVSKI ET AL., 2006, fig.4.31). Further
investigations are required to decide whether this is caused by the drained test condition.
Overall, the result is in good agreement with the fact, that a pronounced anisotropy of the in-situ stress
field is found in the Opalinus clay at Mont Terri with maximum deviatoric stresses between 3 and
7 MPa (reported by several investigators, summarized in HEITZMANN & TRIPET, 2003). Any material
which is able to creep tends to release shear stresses and to approach a hydrostatic state. Therefore,
no creep had to be expected at least up to the mentioned level of deviatoric stress.
Tab. 2 Results from the drained creep tests. The slopes dε/dt are calculated by linear
regression. For the sake of clarity, results with very poor correlations (|R| < 0.6) are printed grey.
Yellow shading indicates phases with increasing sample length. For σdev = 13 MPa there are indications of creep in all samples except BLT-14/11.
sample
dε/dt [1/d]
σdev = 5 MPa
R2
RMSE
dε/dt [1/d]
σdev = 8 MPa
R2
RMSE
dε/dt [1/d]
σdev = 10 MPa
σdev = 13 MPa
R2
BLT-14/05
-8.8·10
-8
0.08
3.1·10
BLT-14/11
-3.0·10
-7
0.20
-6
6.3·10
BLT-15/04
-7.2·10
-7
0.63
-6
-6.2·10-7
9.3·10-8
0.79
0.03
6.2·10
BLT-15/08
-2.7·10-8
0.01
-6
2.6·10-6
-6
9.1·10-6
-8.0·10-8
-2.0·10-7
1.4·10-7
2.1·10-7
0.07
0.10
0.09
0.35
5.0·10
-6
4.3·10-6
6.6·10
dε/dt [1/d]
3.75·10-7
-1.4·10-7
4.3·10-7
1.5·10-7
0.74
0.09
0.61
0.40
RMSE
4.6·10
-6
9.6·10
-6
7.0·10
-6
RMSE
R2
1.4·10
-5
7.1·10
-6
3.8·10-6
LT-Exp. Phase 11+12
Page 11
12
20
10
4
BLT-14/05
BLT-14/11
BLT-15/04
BLT-15/08
0
-4
0
deformation [μm]
deformation [10-5]
8
-10
σz - σx [MPa]
-8
15
10
5
BLT-14
BLT-15
temperature [°C]
0
30
25
20
0
100
time [d]
200
Fig. 7 Results from the drained creep tests. The deformations are displayed as running averages
over 21 data points.
Page 12
LT-Exp. Phase 11+12
2
BLT-14/05
4
BLT-14/11
2
deformation [10-5]
deformation [10-5]
0
0
dε/dt = 3.7·10-7 d-1
R2
= 0.74
RMSE = 4.6·10-6
-2
-2
-4
dε/dt = -1.4·10-7 d-1
R2
= 0.09
RMSE = 9.6·10-6
-6
-4
0
20
60
0
BLT-15/04
4
2
0
dε/dt = 4.3·10-7 d-1
R2
= 0.61
RMSE = 7.1·10-6
-2
20
-4
40
time [d]
60
BLT-15/08
4
deformation [10-5]
deformation [10-5]
40
time [d]
2
0
dε/dt = 1.5·10-7 d-1
R2
= 0.40
RMSE = 3.8·10-6
-2
-4
0
20
40
time [d]
60
0
20
40
time [d]
60
Fig. 8 The creep behavior of the four samples during the test phase with σdev = 13 MPa. Data and
linear regression lines are shown. Deformations are given relative to the "calculated initial state" of this
phase (taken from the linear fit). The alignment of the measured data in horizontal lines displays the
impact of the resolution δlmeas of the combination of A/D-converter and displacement transducer.
δlmeas = 10-6 m. corresponds to a deformation resolution of δεmeas = 4.6·10-6.
LT-Exp. Phase 11+12
Page 13
5 COMPLEX STRENGTH TEST
5.1 CONCEPT AND OBJECTIVE
It was the idea of this test to gain as much information as possible about elastic parameters, the onset
of damage (irreversible deformation, dilatancy etc.), the shear strength, and the residual strength from
a single sample. Furthermore, possible pore pressure effects upon these quantities should be investigated. A Karman cell equipped with a very versatile control and data acquisition system was chosen
for this purpose. It allows for the execution of virtually any type of stress-strain-path.
A test concept comprising three sections was developed. Section A is focused on the investigation of
the onset of damage, section B deals with shear strength, and section C analyses residual strength. In
any section investigations are to be done at various minimum principle stresses (i.e. confining pressures).
All loading phases of the complex strength test were carried out under deformation control at a rate of
dε1/dt = 10-7 s-1. Hydrostatic compaction at 5 MPa over 3 days preceded the first test section. This
roughly corresponds to the in situ mean normal stress at the Mont Terri test site.
Strictly undrained conditions are required during the test because the detection of possible pore pressure effects on the investigated material properties are another objective.
5.1.1 Test section A – limit of linear elasticity
Obviously, the onset of damage must be detected very carefully, if it should be investigated repeatedly
at several levels of confining pressure. To avoid damaging the sample too much, thus changing its
properties significantly and impeding further investigations of an "undisturbed sample", a rather strict
and well detectable criterion for the beginning damage is required. Since all sample deformations are
fully reversible within the range of linear elastic behavior, the limit of linear elasticity might be the very
first evidence for incipient damage. Although it is possibly far below any level of relevant material
damage (such as measurable dilatancy or even failure), it nevertheless characterizes the transition to
another deformation regime, either to a non-linear elastic behavior or to an irreversible alteration of
material properties.
2
5
deformation [10-5]
The main challenge of
this concept is the reliable and "near real time"
detection of the deviation
from linear elasticity.
There are several instances that make this
task difficult:
axial stress, cell pressure [MPa]
To investigate the limit of linear elasticity test section A consists of a sequence of loading cycles. Each
cycle comprises three phases as illustrated in Fig. 9. First, the sample is loaded with a constant rate of
deformation. As soon as a deviation of the axial stress from a linear path can be detected the loading
is stopped. The sample
15
4
is unloaded at a rate of
limit of
deformation
linear elastic
dσ1/dt = -0.1 MPa/min. In
cell pressure
bahaviour
axial stress
the third phase the conlinear elastic stress
3
loading phase
fining pressure is inunloading phase
10
creased while keeping
increasing cell pressure
the axial stress constant.
1
0
time
0
Fig. 9 Sketch of loading cycles performed in section A of the complex
strength test.
Page 14
LT-Exp. Phase 11+12
-
The slope of the deviatoric stress vs. deformation relationship will usually depart from the constant Young's modulus continuously, thus to some extent making the determination of the linear
elastic limit a question of measuring accuracy.
-
Due to measurement noise, observations will always fluctuate a bit around an idealized linear
relationship (Fig. 10). This blurs the onset of a systematic deviation from linear elasticity and particularly hampers the development of an algorithm for automatic detection.
-
Any disturbance of the process control or data acquisition can interfere with the detection of the
linear elastic limit. As shown in Fig. 10, abrupt drops ("spikes") in deviatoric stress (-0.5 to
-1.5 MPa, typically lasting 2 minutes) accidentally occurred in the measurements. They are most
likely a result of a minute tilting of the load piston. They obviously do not affect the sample behavior, but sometimes disturb the determination of the linear elastic limit. Spikes were eliminated
from the data by an appropriate filter.
data deviating from
linear elastic behavior
deviatoric stress σ1- σ3 [MPa]
6
5
all data
data used for fit
linear fit
4
"spike" attributable to a
not yet fully understood
maschine artifact
3
2
data affected by sample hysteresis
and imperfectly compensated hysteresis
of load frame deformation
1
-0.2
-0.21
-0.22
-0.23
deformation σ1 [%]
-0.24
-0.25
Fig. 10 Example of a loading phase performed on a BHE-B1 sample to determine the linear elastic
limit. The confining pressure during this loading was σ3 = 19 MPa.
-
Hysteresis effects of the sample interfere with the linear elastic behavior. They usually lead to an
initial slope of the deviatoric stress vs. deformation relationship which is much steeper than the
Young's modulus. The curve will then bend asymptotically towards the linear relationship. Sometimes, this makes it difficult to determine a linear part of the curve unambiguously and may also
lead to an overestimation of the Young's modulus. In extreme cases, the hysteresis effects might
not fade sufficiently until the linear elastic limit is reached.
-
The loading frame of any triaxial apparatus will exhibit an elastic deformation depending on the
axial load. This frame deformation has to be considered as a correction term in the acquisition of
deformation data. In case of the used triaxial apparatus which is designed for axial forces up to
2500 kN the deformation characteristic is distinctly non-linear (Fig. 11). This is particularly true in
the lower load range up to 500 kN which is applied in the investigation of Mont Terri Opalinus clay
LT-Exp. Phase 11+12
Page 15
frame deformation [mm]
material. Any error in the quantiloading
fication of the frame deforma1.2
unloading
tion will result in a non-linear error of the calculated sample de0.8
formation and might conceal
linearity.
A small but distinct hysteresis of
0.4
the frame deformation (Fig. 11)
even aggravates this problem.
Currently, there is no satisfac0
tory correction available for this
0
500
1000
1500
2000
2500
effect. Therefore, an extensive
axial load [kN]
routine of calibration measurements is on the way to gain an Fig. 11 The loading frame deformation of the used triaxial
apparatus exhibits a strong non-linearity as well as a distinct
improved description of the hysteresis.
frame deformation for use in future experiments as well as for
reprocessing of the existing data sets.
Actually it seems to be too difficult to develop an algorithm for an automatic determination of the linear
elastic limit. Consequently, tools are required to aid the operator to recognize in sufficient time when
the loading has exceeded the linear elastic limit. Obviously, it is not possible to accomplish this task by
means of a simple deviatoric stress vs. deformation plot (Fig. 10).
A self-evident approach is to use a dσ1/dε1-plot. Unfortunately, even a moderate noise in a data set
can disturb the calculation of a derivative dramatically (see Fig. 12 for the impact of noise and smoothing on the dσdev/dε1-curve). The calculation of the derivative has to be based on a data set the range of
which ("smoothing window") is significantly bigger than the noise (in both variables!) to overcome this
problem. In return the smoothing effect of a bigger window leads to increasing errors of the calculated
derivative in regions where the dσ1/dε1-curve bends – which is particularly the case at the linear elastic
limit. Furthermore, a broad smoothing will result in a tendency to localize the beginning of a bending
too early (Fig. 12). Finally, a broad smoothing also hampers the goal of near real time detection of the
linear elastic limit since it delays the data availability by a half width of the smoothing window. Altogether, the dσdev/dε1-curve is not a perfect tool for a timely detection of the linear elastic limit. Nevertheless, it turned out to be helpful in some cases, especially when several variably smoothed
dσdev/dε1-curves are plotted simultaneously.
A "reduced stress" plot, i.e. a plot of the deviation of axial stress from a pure linear elastic behavior
(σred(ε1) = σ1(ε1) – σlin(ε1)), turned out to be the most efficient tool for the detection of the linear elastic
limit. As for the dσdev/dε1-plot, smoothing (by running average) is required in case of noisy data. But a
smaller smoothing window is sufficient because there is no calculation of differential quantities
required. Fig. 13 shows the σred(ε1)-plot for the same artificial data set used in Fig. 12. Note that the
largest smoothing window used in Fig. 13 and the smallest used in Fig. 12 are identical. Contrary to
dσdev/dε1-plots, there is a slight tendency to localize the linear elastic limit too late when using
σred(ε1)-plots.
The application of the dσdev/dε1-plot as well as the σred(ε1)-plot on the measured data set of Fig. 10 is
displayed in Fig. 14.
Page 16
LT-Exp. Phase 11+12
25
exact slope
slope, ideal, Δε=3e-6
slope, noisy, Δε=3e-6
2
20
15
1
ideal deviatoric stress
noisy deviatoric stress
deviatoric stress [MPa]
slope [GPa]
1.5
linear elastic dev. stress
0.5
10
window size for
calculation of slopes
Δε=2e-5
Δε=1e-5
Δε=5e-6
Δε=3e-6
0
range of noise
5
3
4
5
deformation [10-5]
6
7
Fig. 12 The impact of data noise and smoothing on a calculated dσ/dε-curve shown by a synthetic
data set. The assumed material behavior is given by E = 20 GPa, a linear elastic limit of εlin = 5·10-5,
and a decreasing stiffness dσ/dε = (1 – (ε – εlin)/6·10-5) E for εlin < ε < 11·10-5. Besides the "ideal" (noiseless) data, a noisy data set is presented. A normal distributed noise with standard deviations δε = 10-7
and δσ = 0.03 MPa was added to the ideal data. A sampling rate of one measurement per deformation
increment of 10-7 is assumed.
Slopes dσ/dε were calculated by linear regression analysis from the data within a smoothing window
with a half width Δε. Slopes were calculated for various sizes of smoothing windows which are represented at the bottom of the plot including the size of the corresponding σ-window Δσ = E Δε.
The dilemma between an increasing systematic deviation from the exact slope for increasing width of
the smoothing window, and an increasing statistical noise for decreasing width of the smoothing window is obvious.
LT-Exp. Phase 11+12
Page 17
0
-0.1
window size for
running average
Δε=3e-6
Δε=1e-6
Δε=5e-7
linear elastic limit
reduced axial stress [MPa]
0.1
range of noise
ideal reduced stress
noisy reduced stress
smoothed, Δε=3e-6
smoothed, Δε=1e-6
smoothed, Δε=5e-7
-0.2
3
4
5
deformation [10-5]
6
7
Fig. 13 For the synthetic data set used in Fig. 12 the "reduced stress" plot σred(ε) = σ1(ε) – σlin(ε) indicates the linear elastic limit more reliably than a dσ/dε-plot. It is particularly less susceptible towards
noisy data. The size of the smoothing windows used for the calculation of running averages is shown
in the same way as in Fig. 12.
Page 18
LT-Exp. Phase 11+12
20
slopes calculated from
Δε=8·10-6
Δε=2.1·10-5
d ( σ1- σ3 ) / d ε1 [GPa]
16
Δε=4.1·10-5
E-modulus
12
8
4
deviation from linearity
found too early
0
0.2
-0.2
5
all data
data used for fit
running average (Δε=6·10-6)
deviatoric stress
-0.4
-0.6
4
3
-0.8
19.92
deviatoric stress σ1-σ3 [MPa]
0
deviation from linear fit starts
"reduced" deviatoric stress
( σ1- σ3 ) obs - ( σ1- σ3 ) fit [MPa]
6
2
19.94
19.96
time [d]
19.98
20
Fig. 14 The dσdev/dε1-plot as well as the σred(ε1)-plot for a loading phase performed on a BHE-B1
sample to determine the linear elastic limit. The confining pressure during this loading was
σ3 = 19 MPa.
LT-Exp. Phase 11+12
Page 19
5.1.2 Test section B – shear strength
Overall test section B is a conventional strength test. It differs only in the attempt to measure the shear
strength of a single sample repeatedly at different confining pressures σ3. The loading of the sample
was therefore stopped immediately as soon as the σ1(ε)-curve became horizontal. Nevertheless, the
problem already mentioned in chapter 5.1.1 arises by far more severe in this test section: The onset of
any type of damage can only be tested once on an "undisturbed sample". It must be expected that
only very few measured peak stresses approximately represent properties of the undamaged material.
After a few loading cycles, the progressive damage of the sample will change its properties gradually
towards the residual properties.
5.1.3 Test section C – residual strength
Test section C is a conventional test of residual strength. This can be repeated for various levels of
confining pressure σ3. No particular problems have to be expected as long as the required deformation
does not become too large and the sample integrity is not destroyed.
5.2 RESULTS
There was no experience with the execution of a triaxial test as complex as the one outlined before.
Particularly the tools and techniques required for timely decisions concerning the termination of loading phases could not be validated before. Hence, a serious risk to significantly damage the sample
ahead of schedule was suspected. At that time there were only rather few well preserved samples
available from the boreholes BLT-14 to BLT-16. Therefore, a "less valuable" sample from borehole
BHE-B1 was chosen for the "risky" first test. The sample was in p-geometry. Its diameter was 100 mm,
the length 200 mm. Although stored in a sealed plastic bag (but not in the new liners shown in Fig. 3)
since 2002, it displays some clear indications of damage by drying such as shrinkage cracks (Fig. 15).
This of course reduces the chance to find any pore pressure effects during the experiment.
The full test program could be carried out within 53 days without relevant problems. During section A a
leakage in the hydraulic system controlling the axial load resulted in an automatic shutdown of the
triaxial apparatus. But since the sample was not damaged the experiment could be continued.
The confining pressure was increased from 1 MPa up to 21 MPa in steps of 1 MPa in section A and B.
Some pressure steps were skipped in section C. Otherwise, the very long loading phases required to
reach the residual strength might have resulted in too large sample deformations that can destroy
sample integrity or damage some sensors located inside the triaxial cell. The complete stress path of
the experiment is shown in Fig. 16. The occurrence of spikes (see chapter 5.1.1) is quite conspicuous
in these plots. However, they did not harm the sample and did not affect the experimental results except for a few cases where a spike occurs very close to the linear elastic limit and interferes with its
detection.
Page 20
LT-Exp. Phase 11+12
Fig. 15 The BHE-B1/001/15 sample used in the complex strength test. Some shrinkage cracks parallel to the bedding indicate desiccation of the material.
LT-Exp. Phase 11+12
Page 21
section A
6
4
2
0
0
0.05
deformation ε1 [%]
0.1
0.15
1
1.5
section B
20
10
0
0
0.5
deformation ε1 [%]
section C
20
10
section A (before shutdown)
section A (after shutdown)
section B
section C
0
0
5
deformation ε1 [%]
10
15
Fig. 16 The stress-deformation-path σdev(ε1) during the three test sections. In plot B and C the area
of the previous plot is indicated by a grey dashed line.
Page 22
LT-Exp. Phase 11+12
5.2.1 Test section A – limit of linear elasticity
The linear elastic limit was found to exhibit a linear dependency from the minimum principle stress σ3
(Fig. 17, Tab. 3):
σlin = (σ1 – σ3)lin = a0 + a1 σ3
where
a0 = 2.60 MPa
and
a1 = 0.1615
These values are surprisingly small – for instance, they are less than 1/6 of the shear strength (see
below). The linear elastic limit will therefore fall clearly below common damage criteria as the dilatancy
limit.
Concerning the effectiveness of the tools used to determine the linear elastic limit a satisfactory (but
not a perfect) success was found. In some cases there is still an element of subjectiveness in the
selection of the data range used to define the linear elastic curve.
(σ1 - σ3)peak = 15.55 MPa + 1.466 σ3
R2 = 0.9969
RMSE = 0.017 MPa
(σ1 - σ3)res = 11.54 MPa + 0.550 σ3
R2 = 0.9980
RMSE = 0.016 MPa
20
linear elastic limit (poor data)
linear elastic limit (lin. fit)
residual strength
res. strength (lin. fit, σ3 < 8.2 MPa)
res. strength (lin. fit, σ3 > 8.2 MPa)
10
(σ1 - σ3)res = 2.38 MPa + 1.658 σ3
R = 0.9978
RMSE = 0.053 MPa
2
peak stress
peak stress (lin. fit, σ3 < 4.5 MPa)
(σ1 - σ3)lin = 2.60 MPa + 0.1615 σ3
R2 = 0.8910
RMSE = 0.127 MPa
0
0
5
10
15
confining pressure σ3 [MPa]
20
Fig. 17 Linear elastic limit, peak stress and residual strength measured on a single BHE-B1 sample
in the complex strength test.
LT-Exp. Phase 11+12
Page 23
Tab. 3 Fitted σdev(σ3)-relationships for the linear elastic limit, peak stress (shear strength) and
residual strength measured on a single BHE-B1 sample in the complex strength test.
basis of fit
R2
RMSE
[MPa]
min(σ3)
[MPa]
max(σ3)
[MPa]
n
1
21
21
0.8910
shear strength
1
4
4
residual strength
1
8
residual strength
9
21
fit σdev = a0 + a1 σ3
a0
[MPa]
a1
0.127
2.60
0.1615
0.9969
0.017
15.55
1.466
6
0.9978
0.053
2.38
1.658
5
0.9980
0.016
11.54
0.550
As the σlin(σ3)-relationship is linear throughout the investigated range, there is no indication of any
pore pressure effect. To understand how pore pressure effects should appear, a brief description of
the theory might be useful:
According to Terzaghi's effective stress theory (TERZAGHI, 1936), the effective stress1) tensor σ' governs the deformation and damage of a material, and it is given by σ' = σ – u·1 (where σ is the total
stress tensor, u is the pore pressure, and 1 denotes the unity tensor). Consequently, a pore pressure
does not affect shear stresses but reduces normal stresses. Any constitutive equations as well as
stability criteria (e.g. the Mohr-Coulomb relationship) have to be written in effective stresses instead of
total stresses.
Thus, a linear stability criterion like σlin = (σ'1 – σ'3)lin = a0 + a1 σ'3 is equivalent to σlin = (σ1 – σ3)lin
= a0 + a1 (σ3 – u). In case that a pore pressure develops due to compaction when σ'3 or
σ'oct = (σ'1 + σ'2 + σ'3)/3 exceeds a certain limit, the σlin(σ3)-curve will be linear only up to the point where
pore pressure starts to develop, and will fall below the linearity beyond this point. This would be a typical appearance of pore pressure effects.
It should be mentioned that there are more elaborate theories of hydro-mechanic coupling (e.g.
SKEMPTON, 1960). But even for those theories that use fundamentally different concepts than
Terzaghi's approach (e.g. FREDLUND & MORGENSTERN, 1977), the predicted pore pressure effects are
essentially similar.
Besides the determination of the linear elastic limit, Young's modulus E in axial direction (i.e. parallel
to the foliation of the Opalinus clay) could be calculated from section A data (Fig. 18). Results from
loading and unloading phases were in good agreement. For σ3 > 5.5 MPa a constant value
E = 8.61±0.13 GPa was found. The increased and very scattered results for σ3 < 5.5 MPa might be an
artifact arising from deficiencies in the compensation of load frame deformation. This question will not
be resolved until the new calibration data will be available.
1)
The term "effective stress" in this context must not be confused with the elsewhere used "effective shear
stress" τ eff = (σ 1 − σ oct )2 + (σ 2 − σ oct )2 + (σ 3 − σ oct )2
Page 24
LT-Exp. Phase 11+12
unloading
loading
15
lg
lg
E- modulus [GPa]
average E-modulus (σ3 > 5.5 MPa)
10
values probably affected
by imperfect compensation
of load frame deformation
5
0
0
Fig. 18
5
10
15
confining pressure σ3 [MPa]
20
Observed Young's modulus of a BHE-B1 sample in the complex strength test.
5.2.2 Test section B – shear strength
The observed peak stresses σpeak = (σ1 – σ3)peak show a linear increase up to a confining pressure of
approx. 4 MPa (Fig. 17). Thereafter the σpeak(σ3)-curve gradually flattens to a plateau and even to a
slight decrease, until it starts to increase again for σ3 > 15 MPa. The initial linear part is described by
(Tab. 3):
σpeak = a0 + a1 σ3
where
a0 = 15.55 MPa
and
a1 = 1.466
Assuming an isotropic material (which of cause is not the case for the Mont Terri Opalinus clay) this
would correspond to a Mohr-Coulomb line with a cohesive shear k = 4.95 MPa and a friction angle
φ = 25.0°. Then the normal vector to the plane of shear failure should form an angle of
β = 45° + φ/2 = 57.5° relative to the direction of the largest principle stress (i.e. the axis of the sample).
Actually, β = 68° was found for the sample (Fig. 19). This discrepancy clearly points out the anisotropic
nature of the Opalinus clay and corresponds to the expected rotation of the shear plane towards the
bedding plane.
The flattening of the σpeak(σ3)-curve beyond σ3 = 4 MPa is most likely attributable to a progressive
damage of the sample during test section B. While a distinct shear plane was formed, the properties of
the sample gradually approached the residual state and its strength falls far below the original shear
strength. This hypothesis is also strengthened by the observation that the slope of the terminal
increase of the σpeak(σ3)-curve is close to the slope of the residual strength found in section C.
The observed σpeak(σ3)-curve can hardly be explained by pore pressure effects. Whereas the flattening
of the curve would agree with pore pressure effects, the slight decrease around σ3 = 13 MPa and the
subsequent increase does not correspond with theoretical postulations. Nevertheless, there is no
basis for a definite decision concerning the occurrence of pore pressure effects during section B, as
they might be concealed by the impact of progressive sample damage.
LT-Exp. Phase 11+12
Page 25
Fig. 19 The BHE-B1/001/15 sample after the complex strength test. Although the shear failure condition was reached for the first time at an axial deformation ε1 = 0.0046, the test was continued up to
ε1 = 0.155 during section C. This resulted in the formation of very distinctive primary and secondary
shearing interfaces. The shear planes are inclined by approx. 22° versus the sample axis.
5.2.3 Test section C – residual strength
The behavior of the sample during section C was rather heterogeneous with respect to the shape of
the σdev(ε)-curves for different confining pressures and the deformations required to reach a constant
residual stress state (Fig. 16). Regardless this somewhat erratic behavior the observed residual
strength values σres = (σ1 – σ3)res display a very clear picture. The σres(σ3)-curve consists of two linear
braches (Fig. 17, Tab. 3):
σres = a0 + a1 σ3
where
a0 = 2.38 MPa
a0 = 11.54 MPa
and
and
a1 = 1.658
a1 = 0.550
for σ3 < 8.26 MPa
for σ3 > 8.26 MPa
The buildup of pore pressure for minimum principle stresses σ3 > 8 MPa is a possible explanation of
this curve. However, without pore pressure measurements during the test, there is no positive affirmation of this interpretation.
Page 26
LT-Exp. Phase 11+12
6 SUMMARY AND PERSPECTIVE
DRAINED CREEP TEST:
•
In drained creep tests no creep could be detected for a deviatoric stress σdev ≤ 10 MPa. At
σdev = 13 MPa indications of creep were found but require further validation. The observed creep
rates dε/dt ≈ 4·10-7 d-1 are clearly below other published creep test results from Mont Terri
Opalinus clay.
•
More creep tests including longer observation time are required to validate the low creep rates.
•
Actually, the samples are still loaded with σdev = 13 MPa at an increased temperature of 50°C to
investigate the influence of temperature. Undrained tests on the samples after saturation are
planned to assess the impact of pore pressure.
COMPLEX STRENGTH TEST:
•
The tested experimental approach turned out to be suitable to determine the linear elastic limit as
well as the residual strength over a wide range of confining pressure on a single sample. A sample in p-geometry was tested.
•
The linear elastic limit can be described by σlin = (σ1 – σ3)lin = 2.60 MPa + 0.1615 σ3. Young's
modulus is determined as E = 8.61±0.13 GPa.
•
The residual strength shows two linear branches which might indicate a pore pressure effect:
σres = 2.38 MPa + 1.658 σ3 for σ3 < 8.26 MPa
σres = 11.54 MPa + 0.550 σ3 for σ3 > 8.26 MPa
•
Due to progressive damage of the sample, peak strength can be determined for only 3 or 4 values of confining pressure. A linear relationship is found: σpeak = 15.55 MPa + 1.466 σ3.
•
The complex strength test will be executed on a number of p- and s-samples. Further enhancement of the test design is intended (wider range of confining pressure, volume measurement,
damage detection by ultrasonic measurements).
•
For the used triaxial apparatus hysteresis of the load frame deformation turned out to be a major
drawback of deformation measurement. Extensive calibration measurements and the development of a suitable hysteresis model will be carried out to overcome this problem.
7 REFERENCES
CZAIKOVSKI, O., WOLTERS, R., DÜSTERLOH, U. AND LUX, K.-H. (2006): Abschlussbericht zum BMWiForschungsvorhaben Laborative und numerische Grundlagenuntersuchungen zur
Übertragbarkeit von Stoffmodellansätzen und EDV-Software für Endlager im Salzgestein auf
Endlager im Tongestein. – Lehrstuhl für Deponietechnik und Geomechanik, Technische
Universität Clausthal, Germany, 270 p.
FREDLUND, D.G. AND MORGENSTERN, N.R. (1977): Stress state variables for unsaturated soils. – ASCE
J. Geotech. Eng. Div. GT5, 103,447-466.
HEITZMANN, P. AND TRIPET, J.-P. (2003): Mont Terri Project – Geology, Palaeohydrology and Stress
Field of the Mont Terri Region. – Reports of the FOWG, Geology Series, No. 4, 92 p, Bern.
LT-Exp. Phase 11+12
Page 27
KLINKENBERG, M., KAUFHOLD, S., DOHRMANN, R. AND SIEGESMUND, S. (2007): Microstructural investigation of Opalinus Clay – Proposal of a carbonate distribution model. – Abstract, 3rd International Meeting: Clays in Natural & Engineered Barriers for Radioactive Waste Confinement,
ANDRA, Lille, France, p. 345.
PEARSON, F.J., ARCOS, D., BATH, A., BOISSON, J.-Y., FERNANDEZ, A., GAEBLER, H.-E., GAUCHER, E.,
GAUTSCHI, A., GRIFFAULT, L., HERNAN, P. AND WABER, H.N. (2003): Geochemistry of Water in
the Opalinus Clay Formation at the Mont Terri Rock Laboratory - Synthesis Report. –
Geological Report No. 5. Swiss National Hydrological and Geological Survey, Ittigen-Berne,
319 p.
SCHNIER, H. (2004): Postdismantling laboratory triaxial strength tests. – Deliverable 8b, WP3/Task 32
Postdismantling rock mechanic analysis – Heater Experiment (HE). Federal Institute for
Geosciences and Natural Resources (BGR), Hannover.
SKEMPTON, A.W. (1960). Effective stress in soils, concrete and rocks. – Proc. Conf. Pore Pressure and
Suction in Soils, 4–16, Butterworth, London.
TERZAGHI, K. (1936): The shearing resistance of saturated soils and the angle between the planes of
shear. – In: CASAGRANDE, A., RUTLEDGE, P.C. AND WATSON, J.D. (Eds.): Proc. 1st Int. Conf.
Soil Mech. Found. Eng. Vol.1,54-56.
THURY, M. AND BOSSART, P. (1999): Results of hydrogeological, chemical and geotechnical
experiments performed in 1996-1997. – Mont Terri Project. Rapport géologique n°23, Bern,
191 p.
8 LIST OF TABLES
Tab. 1
Sample characterization and test layout for the drained creep tests. ................................... 5
Tab. 2
Results from the drained creep tests. The slopes dε/dt are calculated by linear
regression. For the sake of clarity, results with very poor correlations (|R| < 0.6) are
printed grey. Yellow shading indicates phases with increasing sample length. For
σdev = 13 MPa there are indications of creep in all samples except BLT-14/11..................... 10
Tab. 3
Fitted σdev(σ3)-relationships for the linear elastic limit, peak stress (shear strength)
and residual strength measured on a single BHE-B1 sample in the complex strength
test. ....................................................................................................................................... 23
9 LIST OF FIGURES
Fig. 1
Location of the BLT-14 to BLT-18 boreholes drilled during phase 11+12, and of the
BHE-B1 borehole used in the complex strength test. ........................................................... 2
Fig. 2
Core from BLT-17 showing freshly formed axial fractures. There is neither a
slickenside (which would indicate a tectonic fault) nor any trace of gypsum crystals
(which generate within a few days in any crack opened in the EDZ due to the impact
of atmospheric oxygen) on the interface. .............................................................................. 3
Fig. 3
The liners used for transport and storage of core samples. ................................................. 4
Fig. 4
The alignment of cylindrical samples with respect to the bedding for p- and
s-geometry. ........................................................................................................................... 5
Fig. 5
The p-geometry samples BLT-14/05 and BLT-14/11 before the creep test. ........................ 7
Fig. 6
The s-geometry samples BLT-15/04 and BLT-15/08 before the creep test. ........................ 9
Fig. 7
Results from the drained creep tests. The deformations are displayed as running
averages over 21 data points. ............................................................................................. 11
Page 28
LT-Exp. Phase 11+12
Fig. 8
The creep behavior of the four samples during the test phase with σdev = 13 MPa.
Data and linear regression lines are shown. Deformations are given relative to the
"calculated initial state" of this phase (taken from the linear fit). The alignment of the
measured data in horizontal lines displays the impact of the resolution δlmeas of the
combination of A/D-converter and displacement transducer. δlmeas = 10-6 m.
corresponds to a deformation resolution of δεmeas = 4.6·10-6. ............................................... 12
Fig. 9
Sketch of loading cycles performed in section A of the complex strength test. ................... 13
Fig. 10
Example of a loading phase performed on a BHE-B1 sample to determine the linear
elastic limit. The confining pressure during this loading was σ3 = 19 MPa. .......................... 14
Fig. 11
The loading frame deformation of the used triaxial apparatus exhibits a strong nonlinearity as well as a distinct hysteresis. .............................................................................. 15
Fig. 12
The impact of data noise and smoothing on a calculated dσ/dε-curve shown by a
synthetic data set. ................................................................................................................ 16
Fig. 13
For the synthetic data set used in Fig. 12 the "reduced stress" plot σred(ε) = σ1(ε) –
σlin(ε) indicates the linear elastic limit more reliably than a dσ/dε-plot. It is particularly
less susceptible towards noisy data. The size of the smoothing windows used for the
calculation of running averages is shown in the same way as in Fig. 12. ........................... 17
Fig. 14
The dσdev/dε1-plot as well as the σred(ε1)-plot for a loading phase performed on a
BHE-B1 sample to determine the linear elastic limit. The confining pressure during
this loading was σ3 = 19 MPa. .............................................................................................. 18
Fig. 15
The BHE-B1/001/15 sample used in the complex strength test. Some shrinkage
cracks parallel to the bedding indicate desiccation of the material. ..................................... 20
Fig. 16
The stress-deformation-path σdev(ε1) during the three test sections. In plot B and C
the area of the previous plot is indicated by a grey dashed line. ......................................... 21
Fig. 17
Linear elastic limit, peak stress and residual strength measured on a single BHE-B1
sample in the complex strength test. ................................................................................... 22
Fig. 18
Observed Young's modulus of a BHE-B1 sample in the complex strength test. ................. 24
Fig. 19
The BHE-B1/001/15 sample after the complex strength test. Although the shear
failure condition was reached for the first time at an axial deformation ε1 = 0.0046,
the test was continued up to ε1 = 0.155 during section C. This resulted in the
formation of very distinctive primary and secondary shearing interfaces. The shear
planes are inclined by approx. 22° versus the sample axis. ................................................ 25

Mont Terri Project

Transcription

Similar documents

Faithful Breath - Krautrock